Artificial Intelligence Is Used to Predict CNC Machine Spindle Problems

Machine tool manufacturers use artificial intelligence technology to predict when problems occur on the spindle, ensuring that the workshop can better prepare and schedule related maintenance work.

Artificial intelligence is expected to realize the predictability of predictive maintenance of industrial equipment in a true sense. Previously, the author has covered Cosen Saws' cloud-based predictive maintenance applications. The app not only monitors the saw blade life of the company's CNC sawmill, but also predicts the blade failure before it is about to fail. Another similar example is Mazak's artificial intelligence-based spindle health monitoring system (SHMS). Currently, the system has been selectively installed on the HCN Horizontal Processing Center (HMC).

Both systems are the result of a collaboration with the Industrial Artificial Intelligence Center (formerly the Center for Intelligent Maintenance Systems) at Cincinnati University. Founder and Director Dr Jay Lee is dedicated to predictive and health management research to ensure zero-fault performance of industrial equipment.

The Industrial Artificial Intelligence Center consists of Cincinnati University (UC), Michigan University and Missouri University of Science and Technology. Since 2001, the Center has worked with more than 100 international organizations on more than 100 projects, including Toyota, Boeing, Bosch, Caterpillar, GE Aviation, Goodyear, Harley-Davidson and Siemens, with the goal of eliminating the risk of accidental failure of industrial equipment.

Mazak developed the Spindle Health Monitoring System (SHMS) to ensure that the workshop is able to take spindle maintenance measures long before spindles or spindle bearings are damaged, making it easy to schedule repairs and minimize downtime. Joe Sanders, Mazak's process development coordinator, says the main difference between the artificial intelligence-based system developed by the company and other spindle monitoring technologies is that spindle health monitoring systems (SHMS) do not rely on threshold data. Other spindle monitoring techniques work by detecting the vibration frequency of a spindle and alerting it that the spindle is damaged or about to be damaged. The Spindle Health Monitoring System (SHMS) can detect problems months before they occur, allowing sufficient time for spindle repairs or replacements.

Through a year-long series of spindle destructive tests, a large amount of data was collected, and the artificial intelligence spindle neural network self-organization diagram and spindle-specific characteristics of the spindle health monitoring system (SHMS) system were constructed to determine the difference between normal and abnormal vibrations. Algorithms created from organization charts predict how spindles degrade over time (unless a crash occurs) and show the remaining life as a percentage (Figure 1). "This is different from the timetable for estimating spindle life because we don't know the future application of a machine. It may be used for light work, heavy cutting or round-the-clock work or any other task. Joe Sanders explains.

The main components of spindle health monitoring system (SHMS) include several vibration and current sensors, a data acquisition module and an industrial computer for processing spindle health monitoring system (SHMS) algorithms. After installation, the system performs an hour of modeling testing to determine the operating characteristics of a particular spindle. After that, the user can perform periodic 60s tests to compare the data with the spindle model (Figure 2). Mr. Joe Sanders recommends that if the cycle time is longer, test each part after it has been made. But he also points out that this is not possible if the components have a relatively short cycle time.

Spindle Health Monitoring System (SHMS) can be selectively installed on Mazak HCN machines purchased by the company from Sooth CNC, or used to retrofit machines equipped with Matrix CNC. The company plans to add ball screw predictive monitoring to the Spindle Health Monitoring System (SHMS). In fact, the Industrial Artificial Intelligence Center has been collecting data and developing wear prediction algorithms for ball screws. The next goal is to create an effective spindle health monitoring system (SHMS) technology that continuously monitors spindle health and eliminates reliance on 60s testing. The active system can also be used to detect tool wear and automatically lower cutting parameters to prevent any damage, says Joe Sanders.